● Built for · QA Teams

Synthetic users on
real phones.

AI agents exercise your app like real users, on real phones with real device identity plus emulators. Describe the journey in plain language, get pass or fail with a recorded CI session.

Start for free Book a call

MobiShop QA workflow · demo

All demos on YouTube ↗

Used by teams backed by a16z and Y Combinator

01 · The problem

Why mobile QA stays painful.

Scripts break on release

A renamed id or a moved button kills the whole selector suite.

Real journeys cross apps

Signup spans Gmail, an SMS code and your app.

Simulators hide bugs

OEM skins and regional builds behave nothing like a clean simulator.

02 · The fix

A synthetic user on a real phone.

The agent reads the screen and acts on intent, the way a person would, on a real phone with authentic device identity. Write the test in plain language and a renamed id is a non-event.

Plain-language tests that survive a renamed id. The agent finds the element by intent, not by selector.
Real phones and emulators, concurrent. Ephemeral runs of 5 to 30 minutes draw on your plan's pooled device minutes.
Real device and agent identity that passes the checks real apps run. Cross-app signup with real SMS and OTP, push checks, recorded session and screenshots posted to CI.

03 · What you can run

What QA teams run.

Test UX as a persona

A synthetic user runs the whole journey and flags what confused it.

Real signup through Gmail, an SMS code and your screens.

Test push notifications

Trigger a push, tap it, assert the app opens to the right screen.

Run nightly regression

Scheduled journeys on real phones post pass or fail to CI.

04 · Try these

Real tasks, ready to run.

Fresh install Instagram, sign up, verify by SMS, assert the home feed loads.
Log in to Gmail, search a sender, mark important, archive the reply.
Walk the DoorDash checkout, list every payment option, flag any pre-checked upsell.
Play a show on Netflix, enable subtitles, and add it to My List.

05 · Infrastructure

Mobile execution, as infrastructure.

What a headless browser is to web agents, mobilerun is to mobile agents. You build the product. We run the phones behind it.

The device layer

Real phones, network and identity behind one REST API.

Your product on top

Ship lead-gen tools, assistants, RPA. Own the UX and the customer.

Rent, do not build

No hardware to buy. Start free, run up to 100 phones at once self-serve, and as many as you need on Enterprise.

06 · Get started

Two ways in. Live in minutes.

For developers

API, SDK or MCP.

Get a key. Sign in, provision a device, copy your dr_sk_ key.
Install. npm install @mobilerun/sdk, set MOBILERUN_CLOUD_API_KEY.
Run. tasks.run({ deviceId, task: "Fresh install, sign up, verify by SMS, tap the welcome push, assert home loads" }), or add the MCP server for Claude.

Read the quickstart

No code · Mobile VA

The Chat UI.

Open the Assistant. At cloud.mobilerun.ai/assistant.
Add a phone. Use a phone from your plan or connect your own.
Type the job. “Fresh install, sign up, verify by SMS, tap the welcome push, assert home loads” The VA runs it. Save, schedule, get webhooks.

Open the Assistant

07 · Pricing

Pick your plan.

Run tests on pooled plan minutes, and add a receive-only number for SMS codes.

Plan

Developer

Build and test your automations.

$20 / month

Billed monthly

5 concurrent phones Spin up to 5 phones at the same time. Dedicated phones from add-ons are separate.
600 device minutes Powered-on time across all your phones, pooled per month. 600 minutes are 10 hours. Extra minutes cost $0.03 each.
1,500 credits ($15) Pay for agent usage like the Assistant and Task Runner. 1 credit = $0.01. Extra credits cost $0.01 each.
5 GB residential traffic Data routed through residential IPs, pooled across your phones per month. Extra traffic costs $10 per GB.

Choose Developer

Plan

Startup

Production

Run mobile workflows in production.

$99 / month